The Big Processor Guide - Model Numbers
What follows is a description of some of the more important naming conventions that have been introduced over the years. This is not a comprehensive, all-inclusive list. In particular, it is focused mainly on those chips that have been given model numbers: for instance, the Pentium D. Older processors that were often refered to only by their clock speed (and of course manufacturer and series) don't get a lot of coverage because they don't need it.
Examples: Athlon XP 3200+, Athlon 64 3700+
P-ratings have been applied to the Am5x86, K5, Athlon XP, Athlon MP, Athlon 64, and Athlon 64 X2. The underlying concept of P-Ratings is to provide a basis for comparison. The name originally stood for "Pentium Rating". P-ratings have also been refered to (usually with a touch of derision) as "PR Ratings" in reference to the fact that K5s had model numbers beginning with PR, and to the commonly-held belief the numbers are mainly a Public Relations gimmick.
The concept of the P-rating originated with the Am5x86. For this chip they were used differently than for the rest; the Am5x86 performed substantially slower than the Pentium (a restriction of being compatible with most 486 systems), and consequently it was given a P-rating below its actual clockspeed to make comparisons to the Pentium easier.
K5 P-ratings were intended to allow buyers to compare K5 (specifically, 5k86) processors to Intel's Pentium. The 5k86 performed better in many applications than a Pentium at the same clock speed. In order to make their wares more attractive, AMD created P-ratings. Thus they could market the PR133 (actual clock speed: 116 MHz) as being competitive with the Pentium 133 MHz.
AMD briefly flirted with giving the K6 series "Pentium II Ratings", or PR2. However, performance between the two products was so closely comparable as to make it a waste of effort.
P-ratings were revived and relabeled QuantiSpeed with the introduction of the Athlon XP, shortly after the introduction of the Athlon MP. This time around AMD was insistent that the ratings were in fact based on the Athlon Thunderbird, not any Intel product, and that they were based on a battery of benchmarks run internally. The former claim was largely seen as false from the start. The latter claim began to lose credibility late in the Athlon XP series, and lost all credibility with the Athlon 64 X2 series. The numbers simply got too big, and lost their viability as a real tool for comparison either to Intel's offerings or even to other series from AMD. They remained a usable way to compare chips within a series.
Ordering Part Numbers are AMD's own SKUs, used to uniquely identify a chip based on a number of metrics. Every chip AMD makes has an OPN printed on it. The OPN system is relatively involved, so rather than try to explain it here I will refer the reader to some helpful (hopefully) resources:
- AMD Board
- OC Inside
- If you want information on a particular OPN, you can search for it at AMD's site.
Examples: FX-51, FX-60
The "FX" portion of the name is used to reaffirm the series name. The number is a performance rating (higher is better). Odd numbers are used for single-core processors; even numbers are used for dual-core chips.
Examples: TL-50, TL-56
Turion 64 X2 chips have model numbers starting with two letters, then two numbers. The letters are connected to the letters used for regular Turion 64s. TL chips have a Thermal Design power of 35 W or less. The number is a performance rating. Higher is better.
Examples: ML-34, MT-40
Turion 64 chips have model numbers starting with two letters, then two numbers. The letters designate whether or not the chip is considered "low power". ML chips have a Thermal Design power of 35 W, MT chips have a Thermal Design Power of 25 W. The number is a performance rating. Higher is better. The numbers are comparable between ML and MT chips. Numbers ending in 2 or 8 indicate a chip with 512 KB of L2 cache; all others indicate 1 MB.
Although it was not reflected in model numbers, Athlon XP-Ms are broken up into three groups. Higher-speed chips generally reached the higher-power groups first.
|Type||Voltage||Thermal Design Power|
|Desktop Replacement||1.6 to 1.65 V||72 W|
|Mainstream||1.45 to 1.55 V||45 W|
|Low Voltage||1.35 V||35 W|
Although it was not reflected in model numbers, Mobile Athlon XPs are broken up into two groups. Higher-speed chips generally reached the Standard group first.
|Type||Voltage||Thermal Design Power|
|Standard||1.4 to 1.45 V||35 W|
|Balanced||1.25 to 1.3 V||25 W|
Examples: 1210, 8212, 8214HE
Introduced with the Revision F Opterons (Santa Rosa on Socket F and Santa Ana on Socket AM2), this is a slight modification of the 3-digit scheme described below. The only difference is that another digit has been inserted after the first. At present, that digit is a 2, and is probably meant to correspond to the new (DDR2) memory controller used in these processors. 3, denoting a DDR3 controller, is likely next.
12xx Opterons are Santa Ana core; 22xx and 82xx Opterons are Santa Rosa.
These chips also add another possible suffix , SE, for parts with a higher Thermal Design Power.
Examples: 144, 258, 148HE
Original Opteron names have three parts.
The first part is the first digit. Either 1, 2, or 8, this digit indicates the maximum number of such processors that can be used in one system. So a 1xx chip is usable in single-processor systems, a 2xx chip is usable in 2-processor systems, and an 8xx chip is usable in 8-processor systems.
The second part is a two-digit performance rating (higher is better). These numbers are comparable between groups: the 144 and 244 have the same clock speed. They are not comparable between Opterons with 3-digit model numbers and those with 4-digit model numbers.
The third part is an optional suffix. HE indicates a chip with a lower Thermal Design Power; one that draws less electrical power and produces less heat. EE indicates an even lower-power chip.
Examples: W5580, E5520, L5506
This group, based (at least initially) on the Nehalem core, is a bit complex. The model number is composed of a letter and four digits, which mean:
Letter: Indicates Thermal Design Power (TDP) and how many "ticks" in clock speed (133 MHz increments) the processor can go up when using Turbo Boost, as follows:
|Letter||TDP||Turbo Boost max ticks|
First two digits: Indicate that the part is a member of the Xeon 5500 series (dual-socket server/workstation processors based on the Nehalem core).
Third digit: Indicates amount of L3 cache, QPI link speed, memory controller information, and more.
|Third Digit||L3 cache||QPI link speed||Max memory speed (DDR3)||Turbo Boost support||Hyper-Threading support|
|5-8||8 MB||6.4 GT/s||1333 MHz||Yes||Yes|
|2-4||8 MB||5.86 GT/s||1066 MHz||Yes||Yes|
|0||4 MB||4.8 GT/s||800 MHz||No||No|
Fourth digit: Taken together with the third digit, indicates clock speed and therefore performance relative to other parts where the rest of the model number is the same. Higher clock speeds are indicated by a higher value of these two digits. Parts where the third and fourth digits are the same (such as L5506 and E5506) have the same clock speed.
Examples: E6600, T5500, X6800
Core 2 chips follow the basic convention introduced with the Core Duo and Core Solo; a letter (in certain cases, 2 letters) and four numerical digits. The letter identifies the product series, and the 4-digit number is a model identifier.
|E||Core 2 Duo desktop|
|QX||Core 2 Extreme, quad core|
|T||Core 2 Duo laptop|
|X||Core 2 Extreme, dual core|
A four-digit number follows this letter. The first digit is mostly used to denote a chip's L2 cache or Front Side Bus speed (see table below). The second digit is more or less a performance rating; while in certain cases this is approximately comparable between chips with a different letter or leading digit, in general it is not. In future models, the last two digits will be used to denote different features, such as a faster Front Side Bus.
A limited number of Core 2 Duo laptop parts with a G suffix are on the market. These chips have unlocked multipliers, similar to the Core 2 Extreme.
|First Digit||Series||Type||L2 cache||FSB|
|4||Core 2 Duo||Desktop||2 MB||200 MHZ|
|5||Core 2 Duo||Laptop||2 MB||133 MHz or 166 MHz|
|6||Core 2 Duo or Core 2 Extreme||Desktop||2 MB or 4 MB||266 MHz|
|7||Core 2 Duo||Laptop||4 MB||166 MHz|
After their transition to Socket T, and beginning with the Pentium 4 Prescott, Intel introduced a three-digit model number system. In this system, the first digit identifies the series, while the other two are a relative performance rating. The performance ratings can be compared between chips in the same series. 5, 6, 8, and 9 series ratings correspond insofar as the same center digit means the same clock speed. In most cases, the last digit is used to describe features the chip has: whether it has Hyper-Threading, which core it is based on, and so forth.
The 3 series is odd man out here. In the 3 series, both of the last 2 digits are a performance rating, and they do not directly correspond to the ratings in the other series.
|3||Celeron D (Prescott)|
|5||Pentium 4 (Prescott)|
|6||Pentium 4 (Prescott 2M, Cedar Mill)|
|8||Pentium D and Pentium EE (Smithfield)|
|9||Pentium D and Pentium EE (Presler)|
All are of the form 6x1 and 6x3.
The 662 and 672 (only) support Vanderpool virtualization. All others are named 6x0.
On Socket T
Those named in the form 5xxJ support the No Execute bit. 5x0 and 5x0J support Hyper-Threading. 5x1 support Hyper-Threading, the No Execute bit, and the full set of EM64T extensions. The 5x1 chips were originally refered to as the "F" series when they were being made available to OEMs only. 5x5 and 5x9 do not support Hyper-Threading or EM64T, but do support the No Execute bit if their name ends in J, and have a 133 MHz Front Side Bus. All Prescotts support SSE3 extensions.
Or in short:
|Model||Front Side Bus||Hyper-Threading||No Execute Bit||EM64T|
|5x5 or 5x9||133 MHz||No||No||No|
|5x5J or 5x9J||133 MHz||No||Yes||No|
On Socket 478
The "E" series had a 200 MHz Front Side Bus and support Hyper-Threading. The "A" series had a 133 MHz Front Side Bus and no Hyper-Threading.
"A" Northwoods have a 100 MHz Front Side Bus. "B" Northwoods are 133 MHz. "C" Northwoods are 200 MHz, and these chips also have Hyper-Threading enabled. Of the others, only the 3.06 GHz Northwood "B" has Hyper-Threading enabled.
Examples: Pentium III-S, Pentium III-EB
Some variants of the PIII have a suffix after the series name. On the following table, bold items indicate the feature that was different from the Pentium III based on the same core lacking a suffix.
|B||Katmai||Slot 1||512 KB||133 MHz|
|E||Coppermine||370||256 KB||100 MHz|
|EB||Coppermine||370||256 KB||133 MHz|
|S||Tualatin||370||512 KB||133 MHz|
On Socket 478
No Hyper-Threading, EM64T, or No Execute bit.
On Socket T
Two groups. The first - 3x0J and 3x5J - supported the No Execute bit, did not support Hyper-Threading, and EM64T instructions were present but disabled. The second - 3x1, 3x6, and 355 - activated the EM64T instructions.
Examples: T1300, T2600, L2600
Broken into two groups. Core Solo is single-core. Core Duo is dual-core.
First character is a letter. T for "normal" chips, L for low power ones, U for ultra-low power ones. This is followed by a 4-digit number. The first digit signifies dual or single core (2 for Core Duo, 1 for Core Solo). The next 3 are a performance rating (higher is better). This rating can be compared between Solo and Duo chips (i.e. T1300 and T2300 have the same clock speed), but not between T, L, and U chips, even when the numbers are completely the same.
All chips whose model number begins with U or ends in 50 have a 133 MHz Front Side Bus, while all others have a 166 MHz Front Side Bus. In some cases the xxx50 chips have higher clock speeds than the next-lowest-numbered processor (for instance, the T2250 is 1.73 GHz while the T2300 is 1.66 GHz). This is most likely based on the understanding that higher Front Side Bus speeds (even with lower multipliers) are desirable.
If the model number has an E suffix, that processor does not support Vanderpool virtualization.
Those with names ending in J and all those from the 370 up support the No Execute bit.
Some later-model Xeons have on-die L3 cache to improve performance. For the purposes of this guide, the core name reflects the amount of L3 cache.
Modern Xeons are named with a 4-digit model number. The first digit specifies whether the chip can be used in multi-processor setups (with an arbitrary number of processors) or just dual-processor. 7 for multi-processor, 5 for dual-processor, 3 for single-processor. This also denotes chipset compatibility; processors with a number starting with 5 are compatible with the Intel 5000 chipset series, for instance. The other three are a speed and feature rating, though the methodology behind this is fairly complex. The general guideline is that higher numbers are better.
Examples: E5310, X5355, L5310
Follows the 4-digit scheme above, but with a slight twist. First digit is a 5, second digit is a 3, last two indicate speed. An "E" prefix indicates Thermal Design Power (TDP) of 80 W, an "X" prefix indicates a TDP of 120 W, and an "L" prefix indicates a TDP of 50 W.
Examples: 3070, 3050
Follows the 4-digit scheme above. First digit is a 3, second digit is a 0, last two indicate speed.
Examples: 7110N, 7140M
Follows the 4-digit scheme above. First digit is a 7, second digit is a 1, last two indicate speed. Also has a 1-letter suffix. If suffix is N, chip has a 166 MHz (x4) Front Side Bus; if suffix is M, chip has a 200 MHz (x4) Front Side Bus.
Examples: 5130, 5160
Follows the 4-digit scheme above. First digit is a 5, second digit is a 1, last two indicate speed.
Examples: 5030, 5050
Follows the 4-digit scheme above. First digit is a 5, second digit is a 0, last two indicate speed. There is one MV (Medium Voltage) Xeon Dempsey, the 5063. It has the same Thermal Design Power as the Dempseys with 166 MHz FSB, but has 266 MHz FSB.
In addition to the normal Xeon series, there are some LV Xeons. LV stands for Low Voltage. LV Xeons produce less heat. All Sossaman Xeons are considered LV.
Examples: N270, Z510, Z530P
The first run of Diamondville processors had model numbers starting with either N or Z, and then a 3-digit number to indicated speed. N indicated a 437-ball, 22 mm x 22 mm BGA package that Intel likes to call BGA437, uFCBGA8, or FCBGA8. Z indicated a 441-ball, 13 mm x 14 mm BGA package known variously as BGA441, PBGA441, uSFFBGA, or FCBGA8 USFF.
The second wave muddied the waters a bit by adding some parts with P and PT suffixes. These both indicated that the part used the larger BGA437 package. The two suffixes are different in that "P" parts are rated for the same termperature range as most other Intel processors, while "PT" parts are rated for a wider, "industrial" temperature range. The 3 digits are a consistent performance indicator between different suffixes; the Z510, Z510P, and Z510PT all have the same clock speed and cache size.